def fetch_and_preprocess_irods_metadata_by_metadata(
search_criteria, irods_zone, issues_dict, reference):
"""
This function takes some filtering/matching criteria for selecting data from iRODS based on metadata.
The client also passes an issues_dict to this function as parameter, which the current function just needs to
update with the issues found on the files found in iRODS to match the criteria.
:param issues_dict: an existing dictionary of issues, to which this function needs to add the issues found
:param irods_zone: the irods zone where to search for the data matching the criteria given
:param search_criteria: a dict formed of key= attr name, val = attr value. The operator is by default =.
:return: a dict of key: fpath, value: the iRODS metadata for that path
"""
irods_metadata_by_path = {}
try:
all_files_metadata_objs_list = iRODSMetadataProvider.retrieve_raw_files_metadata_by_metadata(
search_criteria, irods_zone)
except Exception as e:
print(e)
sys.exit(1)
else:
for raw_metadata in all_files_metadata_objs_list:
check_results = []
file_metadata = IrodsSeqFileMetadata.from_raw_metadata(
raw_metadata)
check_results.extend(file_metadata.check_metadata(reference))
irods_metadata_by_path[raw_metadata.fpath] = file_metadata
issues_dict[raw_metadata.fpath].extend(check_results)
return irods_metadata_by_path
def fetch_and_preprocess_irods_metadata_by_path(irods_fpaths, issues_dict,
reference):
"""
This function fetches the irods metadata by file path and preprocesses it.
It also adds the issues found to the issues_dict given as parameter.
:param irods_fpaths:
:param issues_dict:
:param reference:
:return:
"""
irods_metadata_dict = defaultdict(list)
for fpath in irods_fpaths:
try:
raw_metadata = iRODSMetadataProvider.fetch_raw_file_metadata_by_path(
fpath)
except Exception as e:
print(e)
sys.exit(1)
else:
check_results = []
file_metadata = IrodsSeqFileMetadata.from_raw_metadata(
raw_metadata)
check_results.extend(file_metadata.check_metadata(reference))
irods_metadata_dict[fpath] = file_metadata
issues_dict[fpath].extend(check_results)
return irods_metadata_dict
def test_from_raw_metadata_only_replicas(self):
replicas = [
baton_models.DataObjectReplica(number=1, checksum="123abc"),
baton_models.DataObjectReplica(number=2, checksum="abc"),]
raw_metadata = IrodsRawFileMetadata(fpath='/seq/123.bam', file_replicas=replicas)
seq_metadata = IrodsSeqFileMetadata.from_raw_metadata(raw_metadata)
expected = {'name': set(), 'accession_number': set(), 'internal_id': set()}
self.assertEqual(seq_metadata.samples, expected)
self.assertEqual(seq_metadata.libraries, expected)
self.assertEqual(seq_metadata.checksum_in_meta, set())